Clinical Natural Language Processing Research Group

Tools and demos

Our tools and demos for information extraction from clinical texts.

EdIE-Viz: Edinburgh Information Extraction Visualization for Radiology Reports

EdIE-Viz is an interactive natural language processing demo to mine information from brain imaging reports written by radiologists. Given the raw text of a radiology report for a brain scan, the demo creates output for two systems (a rule-based system and a neural-network-based system) that automatically extract medical findings and their locations related to the brain. The tool is also maintained by the Language Technology Group in the School of Informatics.

EdIE-Viz Demo Website

EdIE-Viz on GitHub

Read more in our research: Not a cute stroke: Analysis of Rule- and Neural Network-based Information Extraction Systems for Brain Radiology Reports (Grivas et al., LOUHI, 2020)

Also see our blog post: EdIE, a suite of Information Extraction tools for stroke related phenotypes (Grivas, 2020)

Explainable Automated Medical Coding

An automated medical coding implementation based on deep learning with attention mechanisms to assign ICD codes to discharge summaries. It highlights key words and sentences in a long discharge summary regarding each assigned code.

Explainable-Automated-Medical-Coding on GitHub

Read more in our research: Explainable Automated Coding of Clinical Notes using Hierarchical Label-wise Attention Networks and Label Embedding Initialisation (Dong et al., JBI, 2021)

SemEHR: a Generic Semantic Search System for Electronic Health Records

SemEHR is a generic information extraction and retrieval system to identify contextualised mentions of biomedical concepts from unstructured clinical notes. The system is built upon the off-the-shelf toolkit, Bio-Yodie, and the enterprise search system, CogStack. SemEHR allows an adaptive and iterative IE, by specifying requirements and fine-tuning concepts on a study basis. The tool assembles NLP annotations at the patient level with clinical and EHR-specific knowledge to populate a panorama for each patient, including longitudinal semantic data views and structured medical profile(s). The tool provides ontology-based search and analytics interfaces to retrieve semantic data for clinical studies.

SemEHR on GitHub

Read more in our research: SemEHR: A general-purpose semantic search system to surface semantic data from clinical notes for tailored care, trial recruitment, and clinical research (Wu et al., JAMIA, 2018)